Today, the Bureau of Labor Statistics did what it does every month and released its employment figures. These showed that 51,000 people lost their jobs this month, and that the unemployment rate ticked up to 5.7%, which is a full percentage point higher than it was a year ago. Unsurprisingly, there were losses in both housing-related construction jobs and financial service industry jobs, as well as another 35,000 jobs eliminated in manufacturing. There was also a spike in unemployment for teenagers, now at 20%, whereas the unemployment rate for the general population is actually a not-so-dramatically bad 5.0%.
But these figures are, like all of the figures, are wrong, maybe not spectacularly wrong, but wrong enough that they will get revised over the next three months, then revised again over time, and in the end may look nothing like they do right now.
The household survey generates the unemployment rate and is based on a telephone survey of about 60,000 homes. That captures the self-employed and domestic workers. The payroll survey creates the non-farm employment number and is based on 400,000 businesses (though not all contacted directly -- some are reported from payroll agencies). The payroll survey usually fails to account for start-ups and very small businesses; the payroll survey also count jobs, not people, so someone holding down two payroll jobs is counted twice. There is often a wide discrepancy between the two surveys, and both numbers are subject to numerous seasonal adjustments (for holidays, weather, etc.) and other statistical tools meant to compensate for death rates, retirements, and other factors that can and do alter the size and composition of the work force and hence the ratios and stats that the bureau collects and reports.
For instance, the household survey defines employed and unemployed in the following manner:
"Employment is the total number of employed persons. Included are... self-employed persons, private household workers, agriculture workers, unpaid family workers, and workers on leave without pay during the reference period. Unemployed persons include those who did not have a job during the reference week, had actively looked for work in the prior 4 weeks, and were available for work. Actively looking for work includes activities such as contacting a possible employer, contacting an employment agency or employment center, having a job interview, sending out resumes, filling out job applications, placing or answering job advertisements, and checking union or professional registers. "
Now, if the BLS contacts you on the phone and you are over 15 years old, they will ask you if you have a job; if you say no, they will ask you if you have been looking, and for how long. If you have been looking for work more than 4 weeks or want a job but have stopped looking because you are discouraged, it may surprise you to learn that you are not unemployed; you are, in fact, not in the labor force and hence do not fully exist, at least statistically. If you have looked for a job at some point during the past 12 months but do not currently have one, you may get classified as "marginally attached to the work force."
These statistics are not the work of idiots, however arcane the methods. They are elaborately crafted databases attempting to provide a viable snapshot of a large and fluid population that with the ever-changing nature of work and the decline of massive payrolls is ever-harder to pin down.
However, the statistics are presented as if they were "the truth." Even more disturbing, they become reference points for key decisions and issues. Wall Street seizes on the data as indicators of how stocks should trade and what interest rates should be; economists extrapolate about the health and/or weakness of the system; and politicians and commentators use them to bolster whatever preexisting points they wish to make. In this report, Republicans may well say, hey, unless you're a teenager, the unemployment rate of 5.0% for adults isn't so bad. Democrats may well say, wait a minute, look at urban employment, employment for minorities, and anyone in the housing sector and it looks pretty grim. And back and forth, without any pause to consider that next month and the month after, these very figures will change by tens of thousands and sometimes hundreds of thousands, making the initial discussion almost totally off the mark.
The trends, of course, do matter, and over many moths, we can get a sense of how robust or weak things are. But our desire for certainty and reference points has turned these vague and loose statistics into Holy Grails of certainty. They aren't. That wouldn't matter in and of itself, but they're used for real-world decisions that have real-world implications. If you were told that you were heading north at 50 miles an hour when you wanted to head south at 40 miles an hour, you'd turn around and slow down. But if in fact you were heading west at 38, you'd end up lost or in an accident. That's what we're doing with our stats, and it's a wonder we're still on the road.